Predicting Dynamic Map States from Limited Field-of-View Sensor Data

TL;DR

This paper demonstrates that deep learning models can accurately predict the state of dynamic maps from limited field-of-view sensor data by converting temporal data into a single-image format, enabling effective environment inference.

Contribution

The work introduces a novel approach of representing limited FOV sensor data as a single image for dynamic map prediction, leveraging existing image-to-image models.

Findings

High accuracy in dynamic map state prediction across various sensing scenarios

Effective use of image-to-image models for temporal sensor data

Robustness to occlusions and sensor failures

Abstract

When autonomous systems are deployed in real-world scenarios, sensors are often subject to limited field-of-view (FOV) constraints, either naturally through system design, or through unexpected occlusions or sensor failures. In conditions where a large FOV is unavailable, it is important to be able to infer information about the environment and predict the state of nearby surroundings based on available data to maintain safe and accurate operation. In this work, we explore the effectiveness of deep learning for dynamic map state prediction based on limited FOV time series data. We show that by representing dynamic sensor data in a simple single-image format that captures both spatial and temporal information, we can effectively use a wide variety of existing image-to-image learning models to predict map states with high accuracy in a diverse set of sensing scenarios.